[CP2K-user] [CP2K:21220] SCF convergence problem for high-energy geometries

[Jürg Hutter]

Hi

If you several 'dangling bonds', meaning unpaired electrons, running a closed shell calculation
will not work. Running LSD with a preset multiplicity (most likely high spin) might work.
Another possibility is to run LSD with smearing. In this case the number of alpha and beta electrons
can adjust. However, this option is not compatible with OT.

regards
JH

________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Andrew Qin <qindrew at gmail.com>
Sent: Monday, February 24, 2025 2:15 PM
To: cp2k
Subject: [CP2K:21203] SCF convergence problem for high-energy geometries

Hello,

I want to perform single-point energy and force calculations on a condensed dodecane system (684 atoms, 18 chains) for geometries including a few strained/broken C–C bonds. While the SCF converges for geometries with single broken bond, I encounter convergence issues when multiple bonds (8) are broken. Input and output files for the multiple-bond case are attached.

I am using max_SCF 20 for inner SCF and 70 for outer SCF. After ~3 outer SCF iterations, the energy mostly stabilizes but then drops by 20 Hartrees at iteration 33. Are the any parameters I should change (so far I've tried changing the energy_gap and linesearch values)? Apologies if this is a basic issue—I am new to CP2K and DFT. For context, the energies and forces will be used to train a machine learning potential for polyethylene to capture C–C bond breaking, which is why I use a relatively strict convergence criteria and high grid cutoff.

Thank you!

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/f5696b77-8cc9-4a66-8cb6-fb14f27923cbn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/f5696b77-8cc9-4a66-8cb6-fb14f27923cbn%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB075949090ACEBD66D71DF6A89FC22%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.